False-positive model design matrix, and detection covariates

[northernbio]

Hi: I am doing a study on the consequences of false-positives on occupancy estimates derived from acoustic point counts of songbirds. I am finding misidentification to be an issue, even with very experienced observers. As part of this study I have been simulating false-positive identification errors. Applying Miller et. al's false-positive model has worked quite well. My question is whether, and if so, how I can integrate detection covariates into the false-positive design matrix. Based on some other analyses, I know there are some critical detection covariates that need to be included in the analysis to help correct for false-negatives.

The model I have been using is psi,p(.),p10(.),b(.)



How would I incorporate detection covariates (e.g., recording quality RQ) into the design matrix? Is there documentation available that would help me on this issue?

Rob

[jhines]

You can make any of the detection-related parameters functions of your covariate by entering the covariate name (RQ) into the design matrix. For example, if you want to make the probability that a detection is 'certain', (b parameter), but not the other parameters, the design matrix would look like this:

Code: Select all

          b1    b2    b3     b4
p(1)      1     0      0     0
p(2)      1     0      0     0   
p(3)      1     0      0     0
p(4)      1     0      0     0
p10(1)    0     1      0     0
p10(2)    0     1      0     0
p10(3)    0     1      0     0
p10(4)    0     1      0     0
b(1)      0     0      1     RQ
b(2)      0     0      1     RQ
b(3)      0     0      1     RQ
b(4)      0     0      1     RQ


So, instead of using the 'init' menu to insert the covariate name in all rows of the design matrix, you type the name into each cell.

[northernbio]

Thanks Jim!!